coffee/fast_float
1
0
Fork 0
mirror of https://github.com/fastfloat/fast_float.git synced 2026-09-14 22:52:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

reduce data cache and registers pressure

Browse Source
This commit is contained in:
HedgehogInTheCPP 2026-08-21 17:11:21 +03:00
parent cbdea5e381
commit 8dc88f374a
1 changed files with 19 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

45
include/fast_float/ascii_number.h
View File

@ -401,7 +401,7 @@ loop_parse_if_digits(UC const *&p, UC const *const pend, uint64_t &i) noexcept {
}
// Finalizer
while ((p != pend) && is_integer(*p)) {
i = i * 10 + static_cast<uint64_t>(*p - UC('0')); // may overflow, that's ok
i = i * 10 + static_cast<uint8_t>(*p - UC('0')); // may overflow, that's ok
++p;
}
}
@ -409,20 +409,18 @@ loop_parse_if_digits(UC const *&p, UC const *const pend, uint64_t &i) noexcept {
fastfloat_really_inline FASTFLOAT_CONSTEXPR20 void
loop_parse_if_digits(char const *&p, char const *const pend,
uint64_t &i) noexcept {
#if FASTFLOAT_X86_SIMD >= 42 && FASTFLOAT_USE_SIMD
#if FASTFLOAT_USE_SIMD && FASTFLOAT_X86_SIMD >= 42
if (!is_constant_evaluated()) {
/*
* SSE4.2 handles 16 bytes at once.
*
* This is only used when a complete 16-byte block exists,
* so the load is always inside [p, pend).
*/
while (std::distance(p, pend) >= 16 && parse_if_16_digits(p, i)) {
p += 16;
}
// SSE4.2 handles 16 bytes at once.
while (std::distance(p, pend) >= 16)
if (parse_if_16_digits(p, i)) {
p += 16;
} else {
break;
}
}
#endif
// optimizes better than parse_if_eight_digits_unrolled() for UC = char.
// Optimizes better than parse_if_eight_digits_unrolled() for char.
while (std::distance(p, pend) >= 8 /*sizeof(uint64_t)*/) {
auto const val = read_chars_to_unsigned<uint64_t>(p);
if (is_made_of_8_digits(val)) {
@ -446,7 +444,7 @@ loop_parse_if_digits(char const *&p, char const *const pend,
}
// Finalizer
while ((p != pend) && is_integer(*p)) {
i = i * 10 + static_cast<uint64_t>(*p - '0'); // may overflow, that's ok
i = i * 10 + static_cast<uint8_t>(*p - '0'); // may overflow, that's ok
++p;
}
}
@ -556,32 +554,27 @@ parse_number_string(UC const *p, UC const *pend,
// parser already check that this is num and it's exist
if ((p != pend) && is_integer(*p)) {
#endif
answer.mantissa = static_cast<fast_float::am_mant_t>(*p - UC('0'));
answer.mantissa = static_cast<uint8_t>(*p - UC('0'));
++p;
if ((p != pend) && is_integer(*p)) {
answer.mantissa = static_cast<fast_float::am_mant_t>(
answer.mantissa * 10 +
static_cast<fast_float::am_mant_t>(*p - UC('0')));
answer.mantissa * 10 + static_cast<uint8_t>(*p - UC('0')));
++p;
if ((p != pend) && is_integer(*p)) {
answer.mantissa = static_cast<fast_float::am_mant_t>(
answer.mantissa * 10 +
static_cast<fast_float::am_mant_t>(*p - UC('0')));
answer.mantissa * 10 + static_cast<uint8_t>(*p - UC('0')));
++p;
if ((p != pend) && is_integer(*p)) {
answer.mantissa = static_cast<fast_float::am_mant_t>(
answer.mantissa * 10 +
static_cast<fast_float::am_mant_t>(*p - UC('0')));
answer.mantissa * 10 + static_cast<uint8_t>(*p - UC('0')));
++p;
if ((p != pend) && is_integer(*p)) {
answer.mantissa = static_cast<fast_float::am_mant_t>(
answer.mantissa * 10 +
static_cast<fast_float::am_mant_t>(*p - UC('0')));
answer.mantissa * 10 + static_cast<uint8_t>(*p - UC('0')));
++p;
while ((p != pend) && is_integer(*p)) {
answer.mantissa = static_cast<fast_float::am_mant_t>(
answer.mantissa * 10 +
static_cast<fast_float::am_mant_t>(*p - UC('0')));
answer.mantissa * 10 + static_cast<uint8_t>(*p - UC('0')));
++p;
}
}
@ -694,7 +687,7 @@ parse_number_string(UC const *p, UC const *pend,
while ((p != pend) && is_integer(*p)) {
if (exp_number < am_bias_limit) {
// check for exponent overflow if we have too many digits.
auto const digit = uint8_t(*p - UC('0'));
auto const digit = static_cast<uint8_t>(*p - UC('0'));
exp_number = 10 * exp_number + digit;
}
++p;
@ -906,7 +899,7 @@ parse_int_string(UC const *p, UC const *pend, T &value,
if (is_made_of_4_digits(digits)) {
auto v = parse_4_digits(digits);
if (len >= 5 && is_integer(p[4])) {
v = v * 10 + static_cast<uint32_t>(p[4] - '0');
v = v * 10 + static_cast<uint8_t>(p[4] - '0');
if (len >= 6 && is_integer(p[5])) {
const UC *q = p + 5;
while (q != pend && is_integer(*q)) {